[sdiy] There was a crooked man who had a crooked saw...

[David G. Dixon]

I have a small "problem" with my saw-core VCOs: after the integrator resets,
the integrator output takes a finite time to straighten out.  This gives the
final sawtooth output wave a bit of curvature near the top.  The slope is a
bit too steep for the first fraction of a volt, and then it straightens out.
I've tried everything I can think of to eliminate it, but nothing has any
effect.  It doesn't seem to be related to the comparator RC time constant,
or any other little caps, or the mode of resetting, and it is not reproduced
in SPICE.  One thing I haven't tried is changing the integrator cap.  I'm
using a 1nF, C0G/NP0 ceramic cap.  I have some 1nF metallized polypropylene
film caps I could try, but I haven't yet.

Interestingly enough, the effect more or less disappears at high frequencies
(it is largely undetectable on the scope above the audio range).

Also, the pulse waves derived from the saw have a little downward-pointing
"hook" at the beginning of each negative half-period.  I can't see how the
two phenomena could be related, since the pulse wave is just a comparator
output, and the saw is only setting the threshold voltage.

I'm using all TL07x opamps in the circuit (except for the saw comparator,
which is LM311), and I'm wondering if this has something to do with it.  In
both cases, a TL07x opamp is generating a sharp falling edge.  Could it be
that TL07x opamps do something funny when they generate falling edges?

If anyone has any insight into this phenomenon, I'd love to hear about it.